No-Reference Quality Assessment of Contrast-Distorted Images using Contrast Enhancement
|
|
- Arleen Rice
- 5 years ago
- Views:
Transcription
1 No-Reference Quality Assessment of Contrast-Distorted Images using Contrast Enhancement 1 Jia Yan, Jie Li, Xin Fu Abstract: No-reference image quality assessment (NR-IQA) aims to measure the image quality without reference image. However, contrast distortion has been overlooked in the current research of NR-IQA. In this paper, we propose a very simple but effective metric for predicting quality of contrast-altered images based on the fact that a high-contrast image is often more similar to its contrast enhanced image. Specifically, we first generate an enhanced image through histogram equalization. We then calculate the similarity of the original image and the enhanced one by using structural-similarity index (SSIM) as the first feature. Further, we calculate the histogram based entropy and cross entropy between the original image and the enhanced one respectively, to gain a sum of 4 features. Finally, we learn a regression module to fuse the aforementioned 5 features for inferring the quality score. Experiments on four publicly available databases validate the superiority and efficiency of the proposed technique. Keywords: contrast distortion, no-reference image quality assessment, histogram equalization, similarity measurement, support vector regression 1. Introduction As the users requirements for high-quality images are increasingly rising, no-reference image quality assessment (NR-IQA) has recently become an active research topic, whose objective is to estimate the human perception of image quality without any access to the original reference image [1, 2]. It can be deployed in every location of an image communication system whenever the distorted image is available for evaluation. The majority of current NR-IQA methods are proposed based on two steps, namely, feature extraction and SVR-based regression module. Natural scene statistics (NSS) based features constitute the majority of existing NR-IQA methods. A plethora of NSS models in different spatial or transform domains have been incorporated into state-of-the-art methods [3-12]. In [12], a bag of feature maps over several transform domains and three color spaces are used, which can be seemed as a combination of the previous methods. Further, a growing body of researchers turn to resorting to advanced neural networks and learning systems, e.g., deep belief net [13, 14], shallow Convolutional Neural Networks (CNN) [15, 16] and deep CNNs [17]. However, traditional NR-IQA studies are mainly devoted to gauging commonly encountered artifacts, for example, Gaussian blur, noise, and compression. In comparison, little investigation is dedicated to the issue of NR-IQA for evaluation of the contrast-changed images. It is worth noting that contrast distortion is often introduced during the image acquisition due to limitation of the acquisition device or poor lighting condition. Moreover, while the aforementioned NR-IQA methods work quite well on commonly seen distortions, they always perform ineffectively on contrast distortion [9], as will
2 2 also be shown later in this paper. As a matter of fact, IQA of contrast-distorted images is an important research topic, which can be used as a guider or evaluation criteria in contrast enhancement technologies [18, 19]. Currently, there are several studies only focusing on contrast distorted images. Gu et al. [19, 20] first systematically studied this issue; and designed the CID2013 and CCID2014 databases dedicated to image contrast change, and meanwhile proposed reduced reference IQA (RR-IQA) techniques based on phase congruency and information statistics of the image histogram. Liu et al. [21] also proposed a RR- IQA metric RCIQM with the integration of free energy theory and the histogram comparison of the contrast-altered image with that of the reference image. But these methods rely on partial access to the reference image, and therefore, NR algorithms are eagerly required. Fang et al. [22] proposed a NR quality metric based on the NSS regulation including the moment and entropy features, and it achieves better results than the general-purpose NSS based methods. Gu et al. [23] proposed a training-free blind quality method NIQMC based on the concept of information maximization by computing entropy of particular areas and also comparing the image histogram with the uniformly distributed histogram. Gu et al. [24] lately extracted 17 features based on contrast, sharpness, brightness and more, and then yields an IQA metric BIQME using a regression module, which is learned with gathered big-data training samples. In this paper, we develop a simple yet effective NR-IQA framework for contrast-distorted images by finding that a high contrast image always looks more similar to its enhanced version. Based on this concern, the design philosophy of our blind metric lies in two factors, namely, image similarity, histogram based entropy. Given the input image and the one created through classical histogram equalization (HE), we first concentrate on their appearance. In fact, it is exactly a full-reference IQA problem, and we use classical structural similarity index (SSIM) [25] as the metric. On the other hand, on finding that the histograms of high-contrast and low-contrast images are different, we turn to histograms of the above two images and compute the entropy of original histogram and the enhanced one respectively to reflect the distribution status. Further, we compute the cross entropy to include the interaction between two histograms. Afterward, with the total 5 features, we develop the contrast enhancement based contrast-changed image quality measure (CEIQ) via Support vector regression. Actually, the entropy has already been used for IQA in RCIQM [21] and NIQMC [23]. It is worthwhile to note that CEIQ is different from the previous work. First, the metric RCIQM is not a NR method and needs the histogram of the reference image to compare, while as a NR method, NIQMC compares the input image histogram with the uniformly distributed one. Second, we take advantage of SSIM, which is a benchmark FR-IQA method, in our NR-IQA metric CEIQ because we convert the NR-IQA problem to be a FR-IQA problem by creating a reference image via contrast enhancement. At last, we provide a general framework and any similarity measurement function or contrast enhancement method can be employed in it. The remainder of this paper is organized as follows. In Section II, we present in details the CEIQ method and demonstrate the high relevance of similarities with image contrast. In Section III, thorough experiments verify the superiority and efficiency of our metric. Section IV concludes this paper.
3 3 2. Similarity measurement with equalization The feature extraction flowchart of the proposed CEIQ is illustrated in Fig. 1. It consists of the following main steps: decolorization, histogram equalization, image similarity, histogram based entropy and cross entropy computation, which are described as follows. Given an input color image I, on the first step we convert it to a grayscale image I g for subsequent processing. In particular, the concern here is the choice of the decolorization method. Recently, many methods of color-to-grayscale conversion have been proposed [27-30]. However, most of these methods aim to maximally preserve or even enhance the original color contrast [31]. In our CEIQ framework, our goal is to evaluate the real contrast quality, not to produce the visually-pleasing grayscale result. With this, the original contrast of I should be maintained after decolorization. We produce the grayscale image I g by YIQ color model: I g = I red I green Iblue (1) where I red, I green and I blue denote the RGB channels of I. The YIQ model uses three components: luminance Y and two values I and Q characterizing the chromaticity. This model is used in the American NTSC television norm, and we take the Y channel as the grayscale representation. It should be noted here that the simple method defined in Eq. 1 could not achieve the best grayscale image, but the result image can preserve the original contrast quality with respect to the color image, which is exactly our goal in CEIQ framework. Fig. 2 shows four color images with different contrast quality level selected from CCID2014 database [19], and the quality increases from left to right. Our results using Eq. 1 are shown in the second column. The results via [29] and [30] which are two famous methods for producing visually-pleasing grayscale images, are shown in the third and fourth rows. It bears out the fact that after decolorization, our results reflect very well contrast orders presented in the original images and the gap between the first and fourth images is large, while the quality levels are less discriminable in Fig. 2(c) and Fig. 2(d). The basis of the proposed method is on the assumption that a high-contrast image is often more similar to its result after contrast enhancement. Our concern here is the choice of contrast enhancement method. Generally, the mainstream enhancement methods can be classified into two categories: histogram-based and Retinex-based [32, 33]. In this paper, the simplicity and automation (means it performs well on almost all types of images without parameter tuning) are two factors to be considered. One key issue is then which method could be used in designing CEIQ? In spite of excessive contrast enhancement, histogram equalization (HE) has proved to be a simple and effective image contrast enhancement technique by flattening and stretching the dynamic range of the image s histogram [32]. With this, we take advantage of the classical HE to achieve the contrast enhancement. To illustrate the results of HE, some original color images I c with different contrast, their grayscale images I g and HE results I e are shown in Fig. 3 (a-c). As stated above, in our CEIQ framework, any improvement of conventional HE based method can also be used. Unlike the contrast enhancement methods which only focus on the appearance of the result image, we care both the original and result images, more
4 4 specifically, the similarity of them. Therefore, given two images I g and I e, we continue to extract our contrast-aware features in two aspects. 2.1 Image similarity Intuitively, the similarity functions used in many existing FR-IQA methods can all be used to compute inter-image similarity. In this paper, we adopt the SSIM, which is calculated by: α β λ SSIM( I, I ) = L( I, I ) * C( I, I ) * S( I, I ) (2) g e g e g e g e where constants α, β and λ mediate the relative importance of the three components [25]. L, C and S measure the similarities of luminance, contrast and structure respectively between I g and I e. In this work, we follow [25] for the configuration of the parameters. Therefore, we use the similarity of two images defined in Eq. 2 as the first feature S ge in our CEIQ framework. In our work, we also consider the feature-similarity (FSIM) index [34], which is another excellent FR-IQA method based on SSIM. Its performance will be examined in the section of experimental results. For the five images shown in Fig. 3, the corresponding mean opinion scores (MOS) increase from left to right, which means that the right one has the highest subjective score given by human. Using Eq. 2, we can easily obtain the corresponding s ge scores: , , , and respectively, which are illustrated in Fig. 4. It can be realized that MOS score. Further, we exhibit the scatter plots of MOS/DMOS versus s ge is monotonically related to the s ge on the whole CCID2014 (including 655 images) and CSIQ (including 116 contrast changed images) database [34] in Fig. 5. Clearly, the score of image similarity shows higher linearity and monotonicity on CCID2014 database, but the performance becomes worse on CSIQ database. So, only one feature is not enough to handle all the contrast distortion models and levels, we should explore more contrast-aware features. 2.2 Histogram based entropy and cross entropy The second consideration of our proposed metric comes from the histogram. The major reason is that histogram is the indicator of the contrast. We plot the corresponding histograms in Fig. 3(d) and Fig. 3(e). More precisely, entropy is a chief concept in statistics [26], and it represents the amount of information for a random signal by quantifying its average unpredictability. It is reasonable to assume that not only the high-contrast image, but also its contrast enhanced version are of the histogram toward uniform distribution, which stands for the large entropy. Given the histogram h of pixel values, the entropy E is defined by b E= hi ()log hi () (3) i= 0
5 5 where b indicates the value of bins in histogram and log( ) has base of two. Therefore, given two histograms h g and he corresponding to the input gray image and the equalized image, we can easily obtain their entropy E g and Ee by Eq. 3 as our second and third features. Further, in our work, we take the relationship between h g and he into consideration by the use of the cross entropy, which is computed by: E = h ()log i h () i (4) ge b i= 0 It should be noted here that the combination of Eq. 3 and Eq. 4 is exactly the classical forma of K-L divergence to estimate the distance between two distributions. However, we take the individual part as one feature in our work to describe the characteristic of one histogram and the interaction between two histograms separately. In the same way, another cross entropy is computed by: Thus, we obtain our third and forth features. eg b i= 0 g e e E = h ()log i h () i (5) Finding a good tradeoff between efficacy and efficiency, we divide the histogram into 128 bins when computing, and we find that the value of bins (such as 64, 128 or 256) has little effect on the performance of CEIQ. In addition, we only take the bins with non-zero values to compute the above four features. In Fig. 6, two scatter plots of DMOS versus E e and g Eeg on the CSIQ (including 116 contrast changed images) database are given. From them, we can see that these features perform well than the image similarity on CSIQ database. So, we use the above described five features to represent the contrast quality of the input image. 2.3 The CEIQ metric For any image, we utilize the image similarity s ge, histogram based entropy E g and E e, cross entropy E ge and E eg to form the feature vector f. We can construct a training set of k images with their feature vectors and subjective scores {( f, Q ),( f, Q ),,( f, Q )} A regression function could be learned to find the mapping function. The framework is generic enough to allow for the use of any regressor. In our implementation, a SVM regressor (SVR) is adopted. SVR has previously been applied to IQA problems [4, 7, 10, 11, 12, 20, 22, 24]. In this work, we use the LIBSVM package [36] with the linear kernel to implement the SVR. Once the regression model is learned, we use it to estimate the quality of any input image. k k
6 6 3. Experimental Results and Discussions 3.1 Databases and methods for comparison The source code of our CEIQ has been released at The proposed method is examined on four contrast distortion related image databases: CID2013 [20], CCID2014 [19], CSIQ [34] and TID2013 [37]. CID2013 and CCID2014 are specifically designed for contrast distortion, including 400 and 655 contrast-changed images generated from 15 natural images respectively. CSIQ is built for various distortion types, among which there are 116 global contrast changed images. TID2013 is a large scale image quality database which contains 3000 images generated from 25 reference with 24 distortion types for each reference. A total number of 240 images which are directly related to contrast distortions (#16 mean shift and #17 contrast change) are utilized here. We compare the proposed CEIQ with eight state-of-the-art IQA methods: 1) classical FR-IQA methods, including SSIM [25], FSIM [34] and MAD [38]; 2) representative general-purpose NR-IQA methods, including BRISQUE [4] and IL-NIQE [9]; 3) contrast distortion specific NR-IQA methods, including Fang [22], NIQMC [23] and BIQME [24]. It should be stressed here that BRISQUE, FANG, BIQME and our CEIQ are the methods need training, and we put them into a group. Specially, we also consider the score of s ge as one metric to evaluate its performance, and it should be stressed here that this is a training-free metric. The performance of the involved methods is evaluated by three popular criteria computed between predicted scores and subjective scores: Spearman rank order correlation coefficient (SROCC) and Kendall rank order correlation coefficient (KROCC) for prediction monotonicity, Pearson linear correlation coefficient (PLCC) for prediction accuracy. PLCC is computed after the monotonic logistic mapping between objective and subjective scores [39]. A value close to 1 for SROCC, KROCC and PLCC means superior correlation with subjective opinions. 3.2 Overall performance comparison For the training-free methods, we use all the images in each database to obtain the results. Taking the TID2013 for example, 250 images are all used to compute the coefficients. For the methods require training, we split each database into train and test sets with 80% for training and 20% for testing. The split is conducted according to reference images to guarantee the independency of the image content in training set and test set. This splitting is repeated for 1,000 times and the median results are used to evaluate the final performance. This treatment method is also widely used in training based IQA research, like the experiments in [4, 22, 24]. Table I lists the SROCC, PLCC and KROCC values of examined methods on four databases to show their performance. For better comparing, we divide the NR methods into two groups: training-free and training based. From the results, five conclusions can be derived. First, among training-free methods, even FR ones, our s ge achieves promising results compared to other methods on CID2013, CCID2014 and TID2013 databases. This demonstrate that the image similarity of the input image and the equalized one is really a powerful feature. Second, our CEIQ metric is obviously superior to all the training based models on CCID 2014, CSIQ and TID2013 databases, especially on CSIQ and TID2013 where the CEIQ outperforms others by a significant margin. As to the CID2013 database, the CEIQ is comparable with the best one BIQME ( vs with SROCC). Third, apparently, the contrast distortion specific methods outperform the general-purpose methods (IL-NIQE and BRISQUE) by a large extent
7 7 because of the special selection of contrast-aware features. Fourth, our CEIQ performs more accurately and consistently than our s ge on all the databases. There are two reasons: 1) entropy based features are added for training step means the model can study the difference between distortion levels; 2) the smaller images are included in the evaluation (20% vs. 100%). Fifth, we surprisingly find that our s ge, which is a training-free metric, even outperforms some training based methods on some databases. For example, the SROCC of sge on CID2013 is , while the result of FANG is The results of s ge on the four databases can also be found at Since our CEIQ is a training based metric, which requires distorted images and their MOS to learn the model, we report the performance of it under different training set sizes. 10%-80% of the images in each database are used for training and the left ones are used for test according to the same method described previously. For each ratio, we compute the median SROCC values among 1000 times on each database, and the results are shown in Fig. 7. This figure witnesses slightly fluctuations from 10% to 80%. In fact, by comparing the results in Table I and Fig. 7, even with only 10% of images for training, the SROCC results of our metric are also quite encouraging compared with other metrics. In the above experiments, the training and test samples are drawn from the same database. It is expected that the BIQA model learned from one database should be applicable to images in other databases. Therefore, we test the generalization capability of our metric by training it on one database, then testing it on other databases. The SROCC score is used for evaluation, and the results are presented in Fig. 8. Note that when trained on CID2013 or CCID2014 and then applied to other database, our metric deliver excellent performance and outperform other metrics which are trained and test in the same database in Table I. However, it performs poorly when trained on CSIQ or TID2013 database and test on CID2013 and CCID2014 database. The major reason is that the number of images and the contrast distortion levels of CID2013 and CCID 2014 are more than them of the CSIQ and TID2013. For example, there are 45 distorted images generated from one reference image on average in CCID2014 database, while only 4 distorted images corresponding to one reference image are available in CSIQ database. 3.3 Runtime measure To test the efficiency which is also important in real-world applications, the runtime of involved NR metrics are listed in the Table II. Results are performed on a notebook with Intel Core i7-6700hq CPU@2.6G Hz and 8GB RAM. The software platform is Matlab R2017a. The computational cost is evaluated with RGB image from CCID2014 database. Clearly, our proposed CEIQ owns highest efficiency, followed by BRISQUE and FANG. Compared to the features extracted in BIQME which performs also well in the previous experiments, the major cost of our metric is just the SSIM computation. Hence, the proposed method is quite efficient, outperforming BIQME by a large margin. 3.4 Similarity measurement strategies As described above, our proposed metric illustrated in Fig. 1 provides a general framework for IQA of contrast-distorted images, and any available method can be employed in it, such as the similarity measurement strategies. Besides the SSIM which we used, in this section, we consider another classical
8 8 measurement for inter-image similarity: FSIM [34]. The SROCC results of these methods on CCID2014 database are tabulated in Table III. From the results, it is possible to see that although the FSIM performs better than SSIM in classical FR-IQA problems, its performance is worse than SSIM in our CEIQ. In FSIM, phase congruency (PC) is used as the primary feature [34]. However, PC is contrast invariant, meaning that the images with different contrast distortion level have almost the same scores for PC, which is the main reason of the worse performance of FSIM in our framework. 4 Conclusion In this paper, we develop a no-reference quality metric CEIQ to assess the quality of contrast changed images using histogram equalization from two aspects: image similarity and histogram based entropy and cross entropy. Experimental results show that the proposed methods are well correlated with subjective evaluations even only few images are used for training, suggesting that the proposed metrics are excellent at handling contrast-distorted images with very low computational cost. Further, CEIQ provides a general framework, and other contrast enhancement methods and image similarity measure metrics can be employed in this framework. Acknowledgments This work is supported by the National Nature Science Foundation of China (No ). References [1] W. Lin, C.C. JayKuo, Perceptual visual quality metrics: a survey, Journal of Visual Communication and Image Representation 22 (4) (2011) [2] R. A. Manap, L. Shao, Non-distortion-specific no-reference image quality assessment: a survey, Information Sciences 301 (2015) [3] H. R. Sheikh, A. C. Bovik, L. Cormack, No-reference quality assessment using natural scene statistics: JPEG2000, IEEE Transactions on Image Processing 14 (11) (2005) [4] A. Mittal, A. K. Moorthy, A. C. Bovik, No-reference image quality assessment in the spatial domain, IEEE Transactions on Image Processing 21 (12) (2012) [5] A. K. Moorthy, A. C. Bovik, Blind image quality assessment: From natural scene statistics to perceptual quality, IEEE Transactions on Image Processing 20 (12) (2011) [6] M. A. Saad, A. C. Bovik, C. Charrier, Blind image quality assessment: A natural scene statistics approach in the DCT domain, IEEE Transactions on Image Processing 21 (8) (2012) [7] Y. Zhang, D. M. Chandler, An algorithm for no-reference image quality assessment based on logderivative statistics of natural scenes (2013) [Online]. Available: [8] A. Mittal, R. Soundararajan, A. C. Bovik, Making a completely blind image quality analyzer, IEEE Signal Processing Letters 20 (3) (2013) [9] L. Zhang, L. Zhang, A. C. Bovik, A feature-enriched completely blind image quality evaluator, IEEE Transactions on Image Processing 24 (8) (2015) [10] D. Lee, K. N. Plataniotis, Toward a no-reference image quality assessment using statistics of perceptual color descriptors, IEEE Transactions on Image Processing 25 (8) (2016) [11] J. Yan, W. Zhang, T. Feng, Blind Image Quality Assessment Based on Natural Redundancy Statistics, in: Proceedings of Asian Conference on Computer Vision, 2016, pp
9 9 [12] D. Ghadiyaram, A. C. Bovik, Perceptual quality prediction on authentically distorted images using a bag of features approach, Journal of Vision 17 (1) (2016). [13] W. Hou, X. Gao, D. Tao, X. Li, Blind image quality assessment via deep learning, IEEE Transactions on Neural Networks & Learning Systems 26 (6) (2017) [14] H. Tang, N. Joshi, A. Kapoor, Blind image quality assessment using semi-supervised rectifier networks, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2014, pp [15] L. Kang, P. Ye, Convolutional neutral networks for no-reference image quality assessment, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2014, pp [16] J. Li, J. Yan, D. Deng, W. Shi, S. Deng, No-reference image quality assessment based on hybrid model, Signal Image & Video Processing 11(6) (2017) [17] S. Bianco, L. Celona, P. Napoletano, R. Schettini, On the Use of Deep Learning for Blind Image Quality Assessment [Online]. Available: [18] Z. Chen, T. Jiang, Y. Tian, Quality assessment for comparing image enhancement algorithms, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2014, pp [19] K. Gu, G. Zhai, W. Lin, M. Liu, The Analysis of Image Contrast: From Quality Assessment to Automatic Enhancement, IEEE Transactions on Cybernetics 46 (1) (2015) [20] K. Gu, G. Zhai, X. Yang, W. Zhang, M. Liu, Subjective and objective quality assessment for images with contrast change, in: Proceedings of the 20th IEEE Conference on Image Processing, 2013, pp [21] M. Liu, K. Gu, G. Zhai, J. Zhou, W. Lin, Quality assessment of contrast-altered images, in: Proceedings of the IEEE International Symposium on Circuits and Systems, 2016, pp [22] Y. Fang, K. Ma, Z. Wang, W. Lin, Z. Fang, G. Zhai, No-reference quality assessment of contrastdistorted images based on natural scene statistics, IEEE Signal Processing Letters 22 (7) (2015) [23] K. Gu, W. Lin, G. Zhai, X. Yang, No-Reference Quality Metric of Contrast-Distorted Images Based on Information Maximization, IEEE Transactions on Cybernetics 47 (12) (2017) [24] K. Gu, D. Tao, J. Qiao, W. Lin, Learning a No-Reference Quality Assessment Model of Enhanced Images with Big Data, IEEE Transactions on Neural Networks & Learning Systems PP (99) (2017) [25] Z. Wang, A. C. Bovik, H. R. Sheikh, E. P. Simoncelli, Image quality assessment: From error visibility to structural similarity, IEEE Transactions on Image Processing 13 (4) (2004) [26] C. E. Shannon, A mathematical theory of communication, Bell Syst.Tech. J. 27 (1948) [27] M.D. Fairchild, Color Appearance Models, John Wiley & Sons, [28] R. Bala, R. Eschbach, Spatial Color-to-Grayscale Transform Preserving Chrominance Edge Information, in: Proceedings of the Color and Imaging Conference, 2004, pp [29] M. Grundland, N. A. Dodgson, Decolorize: Fast, contrast enhancing, color to grayscale conversion, Pattern Recognition 40 (11) (2007) [30] C. Lu, L. Xu, J. Jia, Contrast preserving decolorization, in: Proceedings of the IEEE International Conference on Computational Photography, 2012, pp [31] K. Ma, T. Zhao, K. Zeng, Z. Wang, Objective Quality Assessment for Color-to-Gray Image Conversion, IEEE Transactions on Image Processing 24 (12) (2015) [32] M. Kaur, J. Kaur, J. Kaur, Survey of Contrast Enhancement Techniques based on Histogram Equalization, International Journal of Advanced Computer Science & Applications 2 (7) (2011) [33] H. Yue, J. Yang, X. Sun, F. Wu, C. Hou, Contrast Enhancement Based on Intrinsic Image Decomposition, IEEE Transactions on Image Processing 26 (8) (2017) [34] L. Zhang, L. Zhang, X. Mou, D. Zhang, FSIM: a feature similarity index for image quality assessment, IEEE Transactions on Image Processing 20 (8) (2011)
10 10 [35] E. C. Larson, D. M. Chandler, Most apparent distortion: Full reference image quality assessment and the role of strategy, Journal of Electronic Imaging, 19 (1) (2010) [36] C. C. Chang, C. J. Lin, LIBSVM: A Library for Support Vector Machines, [37] N. Ponomarenko, O. Ieremeiev, V. Lukin, K. Egiazarian, L. Jin, J. Astola, B. Vozel, K. Chehdi, M. Carli, F. Battisti, C.C.J. Kuo, Color image database TID2013: Peculiarities and preliminary results, in: Proceedings of the European Workshop on Visual Information Processing, 2013, pp [38] E. C. Larson, D. M. Chandler, Most apparent distortion: Full reference image quality assessment and the role of strategy, Journal of Electronic Imaging 19 (1) (2010 ) [39] VQEG, Final report from the video quality experts group on the validation of objective models of video quality assessment, VQEG, 2000.
11 SSIM Decolorization Sge Input image Histogram equalization CEIQ Histogram computation Entropy Cross Entropy Eg,Ee Ege,Eeg Histogram computation Fig. 1. The feature extraction flowchart of our proposed CEIQ algorithm MOS= MOS= (a) (b) MOS= Color images Grayscale images generated by Eq. 1 c Grayscale images generated by Decolorize [10] d Grayscale images generated by CPD [11] Fig. 2. Comparison with other decolorization methods (a) Color images I b Grayscale images Ig MOS=3.4956
12 (c)contrast enhanced images I e (d) Histogram h g of I g (e) Histogram h e of I e Fig. 3. The histograms of grayscale images and the contrast enhanced ones using HE. Fig. 4. Comparison between S ge and subjective MOS of five images from CCID2014 (a) Fig. 5. Scatter plots of MOS/DMOS versus S ge. (a) CCID2014 database (b) CSIQ database (b)
13 (a) Fig. 6. Scatter plots of DMOS versus E e (a) and E eg (b) on CSIQ database (b) SROCC % 20% 30% 40% 50% 60% 70% 80% CID2013 CCID2014 CSIQ TID2013 Fig. 7. SROCC with different split ratios on four image databases SROCC SROCC (a) Trained on CID2013 SROCC SROCC (b) Trained on CCID2014 (c) Trained on CSIQ (d) Trained on TID2013 Fig. 8 Cross-Database performance of our proposed metric
14 TABLE I Performance comparison of 10 IQA methods. The top one for each group is highlighted. Quality Metrics Type Training CSIQ (116 images) TID2013 (240 images) SROCC PLCC KROCC SROCC PLCC KROCC SSIM FR No FSIM FR No MAD FR No IL-NIQE NR No NIQMC NR No S ge(pro.) NR No BRISQUE NR Yes FANG NR Yes BIQME NR Yes CEIQ(Pro.) NR Yes Quality Metrics Type Training CID2013 (400 images) CCID2014 (655 images) SROCC PLCC KROCC SROCC PLCC KROCC SSIM FR No FSIM FR No MAD FR No IL-NIQE NR No NIQMC NR No S ge(pro.) NR No BRISQUE NR Yes FANG NR Yes BIQME NR Yes CEIQ(Pro.) NR Yes TABLE II Runtime comparison of evaluated NR metrics Metrics Number of features Time (second/image) IL-NIQE NIQMC BRISQUE FANG BIQME CEIQ
15 TABLE III SROCC results of different combination forms in CEIQ Metrics SROCC CEIQ (SSIM) CEIQ (FSIM)
NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED IMAGES BASED ON LOCAL STATISTICS FEATURES
NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED IMAGES BASED ON LOCAL STATISTICS FEATURES Ismail T. Ahmed 1, 2 and Chen Soong Der 3 1 College of Computer Science and Information
More informationImage Quality Assessment based on Improved Structural SIMilarity
Image Quality Assessment based on Improved Structural SIMilarity Jinjian Wu 1, Fei Qi 2, and Guangming Shi 3 School of Electronic Engineering, Xidian University, Xi an, Shaanxi, 710071, P.R. China 1 jinjian.wu@mail.xidian.edu.cn
More informationIMAGE quality assessment (IQA) is a very important step
1 Efficient No-Reference Quality Assessment and Classification Model for Contrast Distorted Images Hossein Ziaei Nafchi, and Mohamed Cheriet, Senior Member, IEEE arxiv:1804.02554v1 [cs.cv] 7 Apr 2018 Abstract
More informationA COMPARATIVE STUDY OF QUALITY AND CONTENT-BASED SPATIAL POOLING STRATEGIES IN IMAGE QUALITY ASSESSMENT. Dogancan Temel and Ghassan AlRegib
A COMPARATIVE STUDY OF QUALITY AND CONTENT-BASED SPATIAL POOLING STRATEGIES IN IMAGE QUALITY ASSESSMENT Dogancan Temel and Ghassan AlRegib Center for Signal and Information Processing (CSIP) School of
More informationDEEP BLIND IMAGE QUALITY ASSESSMENT
DEEP BLIND IMAGE QUALITY ASSESSMENT BY LEARNING SENSITIVITY MAP Jongyoo Kim, Woojae Kim and Sanghoon Lee ICASSP 2018 Deep Learning and Convolutional Neural Networks (CNNs) SOTA in computer vision & image
More informationEfficient Color Image Quality Assessment Using Gradient Magnitude Similarity Deviation
IJECT Vo l. 8, Is s u e 3, Ju l y - Se p t 2017 ISSN : 2230-7109 (Online) ISSN : 2230-9543 (Print) Efficient Color Image Quality Assessment Using Gradient Magnitude Similarity Deviation 1 Preeti Rani,
More informationObjective Quality Assessment of Screen Content Images by Structure Information
Objective Quality Assessment of Screen Content Images by Structure Information Yuming Fang 1, Jiebin Yan 1, Jiaying Liu 2, Shiqi Wang 3, Qiaohong Li 3, and Zongming Guo 2 1 Jiangxi University of Finance
More informationIMPROVING THE PERFORMANCE OF NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED IMAGES USING NATURAL SCENE STATISTICS
201 Jordanian Journal of Computers and Information Technology (JJCIT), Vol. 3, No. 3, December 2017. IMPROVING THE PERFORMANCE OF NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED
More informationImage Quality Assessment Techniques: An Overview
Image Quality Assessment Techniques: An Overview Shruti Sonawane A. M. Deshpande Department of E&TC Department of E&TC TSSM s BSCOER, Pune, TSSM s BSCOER, Pune, Pune University, Maharashtra, India Pune
More informationBlind Image Quality Assessment on Real Distorted Images using Deep Belief Nets
Blind Image Quality Assessment on Real Distorted Images using Deep Belief Nets Deepti Ghadiyaram The University of Texas at Austin Alan C. Bovik The University of Texas at Austin Abstract We present a
More informationBlind Prediction of Natural Video Quality and H.264 Applications
Proceedings of Seventh International Workshop on Video Processing and Quality Metrics for Consumer Electronics January 30-February 1, 2013, Scottsdale, Arizona 1 Blind Prediction of Natural Video Quality
More informationBLIND IMAGE QUALITY ASSESSMENT WITH LOCAL CONTRAST FEATURES
BLIND IMAGE QUALITY ASSESSMENT WITH LOCAL CONTRAST FEATURES Ganta Kasi Vaibhav, PG Scholar, Department of Electronics and Communication Engineering, University College of Engineering Vizianagaram,JNTUK.
More informationRANKING CONSISTENT RATE: NEW EVALUATION CRITERION ON PAIRWISE SUBJECTIVE EXPERIMENTS. Yeji Shen, Tingting Jiang
RANKING CONSISTENT RATE: NEW EVALUATION CRITERION ON PAIRWISE SUBJECTIVE EXPERIMENTS Yeji Shen, Tingting Jiang National Engineering Laboratory for Video Technology Cooperative Medianet Innovation Center
More informationComparison of No-Reference Image Quality Assessment Machine Learning-based Algorithms on Compressed Images
Comparison of No-Reference Image Quality Assessment Machine Learning-based Algorithms on Compressed Images Christophe Charrier 1 AbdelHakim Saadane 2 Christine Fernandez-Maloigne 3 1 Université de Caen-Basse
More informationLearning without Human Scores for Blind Image Quality Assessment
Learning without Human Scores for Blind Image Quality Assessment Wufeng Xue Inst. of Im. Pro. & Pat. Rec Xi an Jiaotong Univ. xwolfs@hotmail.com Lei Zhang Dept. of Computing The Hong Kong Polytechnic Univ.
More informationA Probabilistic Quality Representation Approach to Deep Blind Image Quality Prediction
A Probabilistic Quality Representation Approach to Deep Blind Image Quality Prediction Hui Zeng, Lei Zhang, Senior Member, IEEE Alan C. Bovik, Fellow, IEEE. arxiv:708.0890v2 [cs.cv] 5 Dec 207 Abstract
More informationInternational Journal of Computer Engineering and Applications, Volume XI, Issue IX, September 17, ISSN
Gopika S 1, D. Malathi 2 1 Department of Computer Science, SRM University, Chennai ABSTRACT: Human society always demands for a tool that helps in analyzing the quality of the visual content getting transferred
More informationNo-reference stereoscopic image-quality metric accounting for left and right similarity map and spatial structure degradation
Loughborough University Institutional Repository No-reference stereoscopic image-quality metric accounting for left and right similarity map and spatial structure degradation This item was submitted to
More informationBlind Predicting Similar Quality Map for Image Quality Assessment
Blind Predicting Similar Quality Map for Image Quality Assessment Da Pan, Ping Shi, Ming Hou, Zefeng Ying, Sizhe Fu, Yuan Zhang Communication University of China No.1 Dingfuzhuang East Street Chaoyang
More informationUnsupervised Feature Learning Framework for No-reference Image Quality Assessment
Unsupervised Feature Learning Framework for No-reference Image Quality Assessment Peng Ye, Jayant Kumar, Le Kang, David Doermann Institute for Advanced Computer Studies University of Maryland, College
More informationSCREEN CONTENT IMAGE QUALITY ASSESSMENT USING EDGE MODEL
SCREEN CONTENT IMAGE QUALITY ASSESSMENT USING EDGE MODEL Zhangkai Ni 1, Lin Ma, Huanqiang Zeng 1,, Canhui Cai 1, and Kai-Kuang Ma 3 1 School of Information Science and Engineering, Huaqiao University,
More informationConvolutional Neural Networks for No-Reference Image Quality Assessment
Convolutional Neural Networks for No-Reference Image Quality Assessment Le Kang 1, Peng Ye 1, Yi Li 2, and David Doermann 1 1 University of Maryland, College Park, MD, USA 2 NICTA and ANU, Canberra, Australia
More informationOn the Use of Deep Learning for Blind Image Quality Assessment
Signal Image Video Processing manuscript No. (will be inserted by the editor) On the Use of Deep Learning for Blind Image Quality Assessment Simone Bianco Luigi Celona Paolo Napoletano Raimondo Schettini
More informationNO REFERENCE IMAGE QUALITY ASSESSMENT BASED ON LOCAL BINARY PATTERN STATISTICS
NO REFERENCE IMAGE QUALITY ASSESSMENT BASED ON LOCAL BINARY PATTERN STATISTICS Min Zhang a, Jin Xie b, Xiangrong Zhou a and Hiroshi Fujita a a Department of Intelligent Image Information, Division of Regeneration
More informationLEARNING QUALITY-AWARE FILTERS FOR NO-REFERENCE IMAGE QUALITY ASSESSMENT
LEARNING QUALITY-AWARE FILTERS FOR NO-REFERENCE IMAGE QUALITY ASSESSMENT Zhongyi Gu, Lin Zhang, Xiaoxu Liu, Hongyu Li School of Software Engineering Tongji University Shanghai 201804, China Jianwei Lu
More informationAn Algorithm for No-Reference Image Quality Assessment Based on Log-Derivative Statistics of Natural Scenes
An Algorithm for No-Reference Image Quality Assessment Based on Log-Derivative Statistics of Natural Scenes Yi Zhang and Damon M. Chandler laboratory of Computational Perception and Image Quality School
More informationDCT-BASED IMAGE QUALITY ASSESSMENT FOR MOBILE SYSTEM. Jeoong Sung Park and Tokunbo Ogunfunmi
DCT-BASED IMAGE QUALITY ASSESSMENT FOR MOBILE SYSTEM Jeoong Sung Park and Tokunbo Ogunfunmi Department of Electrical Engineering Santa Clara University Santa Clara, CA 9553, USA Email: jeoongsung@gmail.com
More informationFROM IMAGE QUALITY TO PATCH QUALITY: AN IMAGE-PATCH MODEL FOR NO-REFERENCE IMAGE QUALITY ASSESSMENT. Wen Heng and Tingting Jiang
FROM IMAGE QUALITY TO PATCH QUALITY: AN IMAGE-PATCH MODEL FOR NO-REFERENCE IMAGE QUALITY ASSESSMENT Wen Heng and Tingting Jiang National Engineering Laboratory for Video Technology, Cooperative Medianet
More informationGradient Magnitude Similarity Deviation: A Highly Efficient Perceptual Image Quality Index
1 Gradient Magnitude Similarity Deviation: A Highly Efficient Perceptual Image Quality Index Wufeng Xue a, Lei Zhang b, Member, IEEE, Xuanqin Mou a, Member, IEEE, Alan C. Bovik c, Fellow, IEEE a Institute
More informationAN IMAGE, before it is displayed to a human, is often
IEEE SIGNAL PROCESSING LETTERS, VOL. 23, NO. 1, JANUARY 2016 65 Decision Fusion for Image Quality Assessment using an Optimization Approach Mariusz Oszust Abstract The proliferation of electronic means
More informationBLIND QUALITY ASSESSMENT OF VIDEOS USING A MODEL OF NATURAL SCENE STATISTICS AND MOTION COHERENCY
BLIND QUALITY ASSESSMENT OF VIDEOS USING A MODEL OF NATURAL SCENE STATISTICS AND MOTION COHERENCY Michele A. Saad The University of Texas at Austin Department of Electrical and Computer Engineering Alan
More informationF-MAD: A Feature-Based Extension of the Most Apparent Distortion Algorithm for Image Quality Assessment
F-MAD: A Feature-Based Etension of the Most Apparent Distortion Algorithm for Image Quality Assessment Punit Singh and Damon M. Chandler Laboratory of Computational Perception and Image Quality, School
More informationNo-reference Image Quality Assessment with Deep Convolutional Neural Networks
No-reference Image Quality Assessment with Deep Convolutional Neural Networks Yuming Li, Lai-Man Po, Litong Feng, Fang Yuan Department of Electronic Engineering, City University of Hong Kong, Kowloon,
More informationIMPROVING THE PERFORMANCE OF NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED IMAGES USING NATURAL SCENE STATISTICS
Jordanian Journal of Computers and Information Technology (JJCIT), Vol. 3, No. 3, December 2017. 201 IMPROVING THE PERFORMANCE OF NO-REFERENCE IMAGE QUALITY ASSESSMENT ALGORITHM FOR CONTRAST-DISTORTED
More informationSVD FILTER BASED MULTISCALE APPROACH FOR IMAGE QUALITY ASSESSMENT. Ashirbani Saha, Gaurav Bhatnagar and Q.M. Jonathan Wu
2012 IEEE International Conference on Multimedia and Expo Workshops SVD FILTER BASED MULTISCALE APPROACH FOR IMAGE QUALITY ASSESSMENT Ashirbani Saha, Gaurav Bhatnagar and Q.M. Jonathan Wu Department of
More informationAccelerate Histogram-Based Contrast Enhancement by Selective Downsampling
Accelerate Histogram-Based Contrast Enhancement by Selective Downsampling Gang Cao 1*, Huawei Tian 2, Lifang Yu 3, Xianglin Huang 1 and Yongbin Wang 1 1 School of Computer Science, Communication University
More informationNOVEL PCA-BASED COLOR-TO-GRAY IMAGE CONVERSION. Ja-Won Seo and Seong Dae Kim
NOVEL PCA-BASED COLOR-TO-GRAY IMAGE CONVERSION Ja-Won Seo and Seong Dae Kim Korea Advanced Institute of Science and Technology (KAIST) Department of Electrical Engineering 21 Daehak-ro, Yuseong-gu, Daejeon
More informationOBJECTIVE IMAGE QUALITY ASSESSMENT WITH SINGULAR VALUE DECOMPOSITION. Manish Narwaria and Weisi Lin
OBJECTIVE IMAGE UALITY ASSESSMENT WITH SINGULAR VALUE DECOMPOSITION Manish Narwaria and Weisi Lin School of Computer Engineering, Nanyang Technological University, Singapore, 639798 Email: {mani008, wslin}@ntu.edu.sg
More informationNo-Refrence Image Quality Assessment Using Blind Image Quality Indices
No-Refrence Image Quality Assessment Using Blind Image Quality Indices Parul Satsangi, Sagar Tandon, Prashant Kr. Yadav & Priyal Diwakar Electronics and Communication Department, M.I.T,Moradabad E-mail
More informationBlind Image Quality Assessment Through Wakeby Statistics Model
Blind Image Quality Assessment Through Wakeby Statistics Model Mohsen Jenadeleh (&) and Mohsen Ebrahimi Moghaddam Faculty of Computer Science and Engineering, Shahid Beheshti University G.C, Tehran, Iran
More information3. Learning from rankings for NR-IQA
RankIQA: Learning from Rankings for No-reference Image Quality Assessment Xialei Liu Computer Vision Center Barcelona, Spain xialei@cvcuabes Joost van de Weijer Computer Vision Center Barcelona, Spain
More informationA Comparison of Still-Image Compression Standards Using Different Image Quality Metrics and Proposed Methods for Improving Lossy Image Quality
A Comparison of Still-Image Compression Standards Using Different Image Quality Metrics and Proposed Methods for Improving Lossy Image Quality Multidimensional DSP Literature Survey Eric Heinen 3/21/08
More informationImage and Video Quality Assessment Using Neural Network and SVM
TSINGHUA SCIENCE AND TECHNOLOGY ISSN 1007-0214 18/19 pp112-116 Volume 13, Number 1, February 2008 Image and Video Quality Assessment Using Neural Network and SVM DING Wenrui (), TONG Yubing (), ZHANG Qishan
More informationColor Local Texture Features Based Face Recognition
Color Local Texture Features Based Face Recognition Priyanka V. Bankar Department of Electronics and Communication Engineering SKN Sinhgad College of Engineering, Korti, Pandharpur, Maharashtra, India
More informationObjective image quality assessment
Hot Topics in Multimedia Research Objective image quality assessment (IQA) aims to develop computational models that can measure image quality consistently in subjective evaluations. With the now widespread
More informationA New Spatial Hue Angle Metric for Perceptual Image Difference
A New Spatial Hue Angle Metric for Perceptual Image Difference Marius Pedersen 1,2 and Jon Yngve Hardeberg 1 1 Gjøvik University College, Gjøvik, Norway 2 Océ Print Logic Technologies S.A., Créteil, France
More informationA PERCEPTUALLY RELEVANT SHEARLET-BASED ADAPTATION OF THE PSNR
A PERCEPTUALLY RELEVANT SHEARLET-BASED ADAPTATION OF THE PSNR Sebastian Bosse, Mischa Siekmann, Wojciech Samek, Member, IEEE, and Thomas Wiegand,2, Fellow, IEEE Fraunhofer Institute for Telecommunications,
More informationReal-time No-Reference Image Quality Assessment based on Filter Learning
2013 IEEE Conference on Computer Vision and Pattern Recognition Real-time No-Reference Image Quality Assessment based on Filter Learning Peng Ye, Jayant Kumar, Le Kang, David Doermann Institute for Advanced
More informationKeywords: Contrast Masking, Gradient Similarity, Human Visual System (HVS), Image Quality Assessment (IQA), Structural Similarity (SSIM).
ISSN 2348 2370 Vol.06,Issue.02, March-2014, Pages:64-71 www.semargroup.org Image Quality Assessment Based on Gradient Similarity F. ASMA BEGUM 1, NAZIA SHABANA 2, NAHID JABEEN 3 1 Assoc Prof, Dept of ECE,
More informationMULTI-SCALE STRUCTURAL SIMILARITY FOR IMAGE QUALITY ASSESSMENT. (Invited Paper)
MULTI-SCALE STRUCTURAL SIMILARITY FOR IMAGE QUALITY ASSESSMENT Zhou Wang 1, Eero P. Simoncelli 1 and Alan C. Bovik 2 (Invited Paper) 1 Center for Neural Sci. and Courant Inst. of Math. Sci., New York Univ.,
More informationA new spatial hue angle metric for perceptual image difference
A new spatial hue angle metric for perceptual image difference Marius Pedersen 1,2 and Jon Yngve Hardeberg 1 1 Gjøvik University College, Gjøvik, Norway 2 Océ Print Logic Technologies S.A., Créteil, France.
More informationA Regression-Based Family of Measures for Full-Reference Image Quality Assessment
Journal homepage: http://www.degruyter.com/view/j/msr A Regression-Based Family of Measures for Full-Reference Image Quality Assessment Mariusz Oszust Department of Computer and Control Engineering, Rzeszow
More informationAn ICA based Approach for Complex Color Scene Text Binarization
An ICA based Approach for Complex Color Scene Text Binarization Siddharth Kherada IIIT-Hyderabad, India siddharth.kherada@research.iiit.ac.in Anoop M. Namboodiri IIIT-Hyderabad, India anoop@iiit.ac.in
More informationDETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT
DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT 1 NUR HALILAH BINTI ISMAIL, 2 SOONG-DER CHEN 1, 2 Department of Graphics and Multimedia, College of Information
More informationGradient Magnitude Similarity Deviation: A Highly Efficient Perceptual Image Quality Index
1 Gradient Magnitude Similarity Deviation: A Highly Efficient Perceptual Image Quality Index Wufeng Xue a,b, Lei Zhang b, Member, IEEE, Xuanqin Mou a, Member, IEEE, Alan C. Bovik c, Fellow, IEEE a Institute
More informationNo-reference image quality assessment based on quality patches in real time
Zhang et al. EURASIP Journal on Image and Video Processing (2018) 2018:111 https://doi.org/10.1186/s13640-018-0361-z EURASIP Journal on Image and Video Processing RESEARCH No-reference image quality assessment
More informationMS-UNIQUE: Multi-model and Sharpness-weighted Unsupervised Image Quality Estimation
MS-UNIQUE: Multi-model and Sharpness-weighted Unsupervised Image Quality Estimation M. Prabhushankar, D.Temel, and G. AlRegib Center for Signal and Information Processing School of Electrical and Computer
More informationEdge-directed Image Interpolation Using Color Gradient Information
Edge-directed Image Interpolation Using Color Gradient Information Andrey Krylov and Andrey Nasonov Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics,
More informationSUBJECTIVE ANALYSIS OF VIDEO QUALITY ON MOBILE DEVICES. Anush K. Moorthy, Lark K. Choi, Gustavo de Veciana and Alan C. Bovik
SUBJECTIVE ANALYSIS OF VIDEO QUALITY ON MOBILE DEVICES Anush K. Moorthy, Lark K. Choi, Gustavo de Veciana and Alan C. Bovik Department of Electrical and Computer Engineering, The University of Texas at
More informationBLIND QUALITY ASSESSMENT OF JPEG2000 COMPRESSED IMAGES USING NATURAL SCENE STATISTICS. Hamid R. Sheikh, Alan C. Bovik and Lawrence Cormack
BLIND QUALITY ASSESSMENT OF JPEG2 COMPRESSED IMAGES USING NATURAL SCENE STATISTICS Hamid R. Sheikh, Alan C. Bovik and Lawrence Cormack Laboratory for Image and Video Engineering, Department of Electrical
More informationReversible Image Data Hiding with Local Adaptive Contrast Enhancement
Reversible Image Data Hiding with Local Adaptive Contrast Enhancement Ruiqi Jiang, Weiming Zhang, Jiajia Xu, Nenghai Yu and Xiaocheng Hu Abstract Recently, a novel reversible data hiding scheme is proposed
More informationLEARNING NATURAL STATISTICS OF BINOCULAR CONTRAST FOR NO REFERENCE QUALITY ASSESSMENT OF STEREOSCOPIC IMAGES
LEARNING NATURAL STATISTICS OF BINOCULAR CONTRAST FOR NO REFERENCE QUALITY ASSESSMENT OF STEREOSCOPIC IMAGES Yi Zhang, Damon M. Chandler Laboratory of Computational Perception and Image Quality Department
More informationA Joint Histogram - 2D Correlation Measure for Incomplete Image Similarity
A Joint Histogram - 2D Correlation for Incomplete Image Similarity Nisreen Ryadh Hamza MSc Candidate, Faculty of Computer Science & Mathematics, University of Kufa, Iraq. ORCID: 0000-0003-0607-3529 Hind
More information2016 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 26, NO. 4, APRIL 2017
2016 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 26, NO. 4, APRIL 2017 Objective Quality Assessment of Screen Content Images by Uncertainty Weighting Yuming Fang, Member, IEEE, Jiebin Yan, Jiaying Liu,
More informationNo-Reference Image Based On Fuzzy Classification for Quality Assessment
No-Reference Image Based On Fuzzy Classification for Quality Assessment Percy Irine Biju P Department of Electrical and Electronics Engineering Anna University Regional Campus, Tirunelveli, Tamilnadu,
More informationA NO-REFERENCE VIDEO QUALITY PREDICTOR FOR COMPRESSION AND SCALING ARTIFACTS
A NO-REFERENCE VIDEO QUALITY PREDICTOR FOR COMPRESSION AND SCALING ARTIFACTS Deepti Ghadiyaram The University of Texas at Austin Chao Chen, Sasi Inguva, and Anil Kokaram Google Inc. ABSTRACT No-Reference
More informationTexture Image Segmentation using FCM
Proceedings of 2012 4th International Conference on Machine Learning and Computing IPCSIT vol. 25 (2012) (2012) IACSIT Press, Singapore Texture Image Segmentation using FCM Kanchan S. Deshmukh + M.G.M
More informationBLIND image quality assessment (BIQA) aims to predict. End-to-End Blind Image Quality Assessment Using Deep Neural Networks
IEEE TRANSACTIONS ON IMAGE PROCESSING 1 End-to-End Blind Image Quality Assessment Using Deep Neural Networks Kede Ma, Student Member, IEEE, Wentao Liu, Student Member, IEEE, Kai Zhang, Zhengfang Duanmu,
More informationIMAGE QUALITY ASSESSMENT BASED ON EDGE
IMAGE QUALITY ASSESSMENT BASED ON EDGE Xuanqin Mou 1, Min Zhang 1, Wufeng Xue 1 and Lei Zhang 1 Institute of Image Processing and Pattern Recognition, Xi'an Jiaotong University, China Department of Computing,
More informationImage Quality Assessment Based on Regions of Interest
Noname manuscript No. (will be inserted by the editor) Image Quality Assessment Based on Regions of Interest A. Alaei R. Raveaux D. Conte Received: date / Accepted: date Abstract Most methods in the literature
More informationIEEE TRANSACTIONS ON IMAGE PROCESSING, A Feature-Enriched Completely Blind Image Quality Evaluator
IEEE TRANSACTIONS ON IMAGE PROCESSING, 205 A Feature-Enriched Completely Blind Image Quality Evaluator Lin Zhang, Member, IEEE, Lei Zhang, Senior Member, IEEE, and Alan C. Bovik, Fellow, IEEE Abstract
More informationBlock Mean Value Based Image Perceptual Hashing for Content Identification
Block Mean Value Based Image Perceptual Hashing for Content Identification Abstract. Image perceptual hashing has been proposed to identify or authenticate image contents in a robust way against distortions
More informationRobust Image Watermarking based on DCT-DWT- SVD Method
Robust Image Watermarking based on DCT-DWT- SVD Sneha Jose Rajesh Cherian Roy, PhD. Sreenesh Shashidharan ABSTRACT Hybrid Image watermarking scheme proposed based on Discrete Cosine Transform (DCT)-Discrete
More informationA DCT Statistics-Based Blind Image Quality Index
A DCT Statistics-Based Blind Image Quality Index Michele Saad, Alan C. Bovik, Christophe Charrier To cite this version: Michele Saad, Alan C. Bovik, Christophe Charrier. A DCT Statistics-Based Blind Image
More informationEdge-Directed Image Interpolation Using Color Gradient Information
Edge-Directed Image Interpolation Using Color Gradient Information Andrey Krylov and Andrey Nasonov Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics,
More informationA Full Reference Based Objective Image Quality Assessment
A Full Reference Based Objective Image Quality Assessment Mayuresh Gulame, K. R. Joshi & Kamthe R. S. P.E.S Modern College of Engineering, Pune -5 E-mail : mayuresh2103@gmail.com, krjpune@gmail.com rupalikamathe@gmail.com
More informationA Fast Video Illumination Enhancement Method Based on Simplified VEC Model
Available online at www.sciencedirect.com Procedia Engineering 29 (2012) 3668 3673 2012 International Workshop on Information and Electronics Engineering (IWIEE) A Fast Video Illumination Enhancement Method
More informationA FRAMEWORK OF EXTRACTING MULTI-SCALE FEATURES USING MULTIPLE CONVOLUTIONAL NEURAL NETWORKS. Kuan-Chuan Peng and Tsuhan Chen
A FRAMEWORK OF EXTRACTING MULTI-SCALE FEATURES USING MULTIPLE CONVOLUTIONAL NEURAL NETWORKS Kuan-Chuan Peng and Tsuhan Chen School of Electrical and Computer Engineering, Cornell University, Ithaca, NY
More informationA reversible data hiding based on adaptive prediction technique and histogram shifting
A reversible data hiding based on adaptive prediction technique and histogram shifting Rui Liu, Rongrong Ni, Yao Zhao Institute of Information Science Beijing Jiaotong University E-mail: rrni@bjtu.edu.cn
More informationA Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation
, pp.162-167 http://dx.doi.org/10.14257/astl.2016.138.33 A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation Liqiang Hu, Chaofeng He Shijiazhuang Tiedao University,
More informationHigh Capacity Reversible Watermarking Scheme for 2D Vector Maps
Scheme for 2D Vector Maps 1 Information Management Department, China National Petroleum Corporation, Beijing, 100007, China E-mail: jxw@petrochina.com.cn Mei Feng Research Institute of Petroleum Exploration
More informationResearch on Design and Application of Computer Database Quality Evaluation Model
Research on Design and Application of Computer Database Quality Evaluation Model Abstract Hong Li, Hui Ge Shihezi Radio and TV University, Shihezi 832000, China Computer data quality evaluation is the
More informationImage Quality Assessment: From Error Visibility to Structural Similarity. Zhou Wang
Image Quality Assessment: From Error Visibility to Structural Similarity Zhou Wang original Image Motivation MSE=0, MSSIM=1 MSE=225, MSSIM=0.949 MSE=225, MSSIM=0.989 MSE=215, MSSIM=0.671 MSE=225, MSSIM=0.688
More informationImage Classification for JPEG Compression
Image Classification for Compression Jevgenij Tichonov Vilnius University, Institute of Mathematics and Informatics Akademijos str. 4 LT-08663, Vilnius jevgenij.tichonov@gmail.com Olga Kurasova Vilnius
More informationAn efficient face recognition algorithm based on multi-kernel regularization learning
Acta Technica 61, No. 4A/2016, 75 84 c 2017 Institute of Thermomechanics CAS, v.v.i. An efficient face recognition algorithm based on multi-kernel regularization learning Bi Rongrong 1 Abstract. A novel
More informationDesign of student information system based on association algorithm and data mining technology. CaiYan, ChenHua
5th International Conference on Mechatronics, Materials, Chemistry and Computer Engineering (ICMMCCE 2017) Design of student information system based on association algorithm and data mining technology
More informationSPATIO-TEMPORAL SSIM INDEX FOR VIDEO QUALITY ASSESSMENT
SPATIO-TEMPORAL SSIM INDEX FOR VIDEO QUALITY ASSESSMENT Yue Wang 1,Tingting Jiang 2, Siwei Ma 2, Wen Gao 2 1 Graduate University of Chinese Academy of Sciences, Beijing, China 2 National Engineering Lab
More informationSTUDYING OF CLASSIFYING CHINESE SMS MESSAGES
STUDYING OF CLASSIFYING CHINESE SMS MESSAGES BASED ON BAYESIAN CLASSIFICATION 1 LI FENG, 2 LI JIGANG 1,2 Computer Science Department, DongHua University, Shanghai, China E-mail: 1 Lifeng@dhu.edu.cn, 2
More informationOptimizing the Deblocking Algorithm for. H.264 Decoder Implementation
Optimizing the Deblocking Algorithm for H.264 Decoder Implementation Ken Kin-Hung Lam Abstract In the emerging H.264 video coding standard, a deblocking/loop filter is required for improving the visual
More informationResearch on Clearance of Aerial Remote Sensing Images Based on Image Fusion
Research on Clearance of Aerial Remote Sensing Images Based on Image Fusion Institute of Oceanographic Instrumentation, Shandong Academy of Sciences Qingdao, 266061, China E-mail:gyygyy1234@163.com Zhigang
More informationBlind Quality Prediction of Stereoscopic 3D Images
Blind Quality Prediction of Stereoscopic 3D Images Jiheng Wang, Qingbo Wu, Abdul Rehman, Shiqi Wang, and Zhou Wang Dept. of Electrical & Computer Engineering, University of Waterloo; Waterloo, ON, Canada
More informationImage Distortion Detection using Convolutional Neural Network
2017 4th IAPR Asian Conference on Pattern Recognition Image Distortion Detection using Convolutional Neural Network Namhyuk Ahn Ajou University Suwon, Korea aa0dfg@ajou.ac.kr Byungkon Kang Ajou University
More informationAN EFFICIENT BATIK IMAGE RETRIEVAL SYSTEM BASED ON COLOR AND TEXTURE FEATURES
AN EFFICIENT BATIK IMAGE RETRIEVAL SYSTEM BASED ON COLOR AND TEXTURE FEATURES 1 RIMA TRI WAHYUNINGRUM, 2 INDAH AGUSTIEN SIRADJUDDIN 1, 2 Department of Informatics Engineering, University of Trunojoyo Madura,
More informationPerformance of Quality Metrics for Compressed Medical Images Through Mean Opinion Score Prediction
RESEARCH ARTICLE Copyright 212 American Scientific Publishers All rights reserved Printed in the United States of America Journal of Medical Imaging and Health Informatics Vol. 2, 1 7, 212 Performance
More informationTemperature Calculation of Pellet Rotary Kiln Based on Texture
Intelligent Control and Automation, 2017, 8, 67-74 http://www.scirp.org/journal/ica ISSN Online: 2153-0661 ISSN Print: 2153-0653 Temperature Calculation of Pellet Rotary Kiln Based on Texture Chunli Lin,
More informationSurvey on Multi-Focus Image Fusion Algorithms
Proceedings of 2014 RAECS UIET Panjab University Chandigarh, 06 08 March, 2014 Survey on Multi-Focus Image Fusion Algorithms Rishu Garg University Inst of Engg & Tech. Panjab University Chandigarh, India
More informationPerceptual Quality Prediction on Authentically Distorted Images Using a Bag of Features Approach
Journal of Vision (26) http://journalofvision.org/ Perceptual Quality Prediction on Authentically Distorted Images Using a Bag of Features Approach Deepti Ghadiyaram Department of Computer Science University
More informationRECENT years have seen a huge growth in the popularity. No-Reference Image Quality Assessment for High Dynamic Range Images
1 No-Reference Image Quality Assessment for High Dynamic Range Images Debarati Kundu, Deepti Ghadiyaram, Alan C. Bovik and Brian L. Evans Wireless Networking and Communications Group Department of Electrical
More informationEvaluation of Two Principal Approaches to Objective Image Quality Assessment
Evaluation of Two Principal Approaches to Objective Image Quality Assessment Martin Čadík, Pavel Slavík Department of Computer Science and Engineering Faculty of Electrical Engineering, Czech Technical
More informationA QUAD-TREE DECOMPOSITION APPROACH TO CARTOON IMAGE COMPRESSION. Yi-Chen Tsai, Ming-Sui Lee, Meiyin Shen and C.-C. Jay Kuo
A QUAD-TREE DECOMPOSITION APPROACH TO CARTOON IMAGE COMPRESSION Yi-Chen Tsai, Ming-Sui Lee, Meiyin Shen and C.-C. Jay Kuo Integrated Media Systems Center and Department of Electrical Engineering University
More information